The SAM (Search of Associative Memory) model, developed by Jeroen Raaijmakers and Richard Shiffrin in 1981, provides a comprehensive mathematical account of human memory retrieval. It formalizes the process by which memory probes activate stored traces and the conditions under which activated traces lead to successful recall.
Sampling and Recovery
SAM distinguishes two stages of retrieval. In the sampling stage, items in memory are activated in proportion to their associative strength to the retrieval cues. The probability of sampling item i is its strength relative to the total strength of all items in the search set. In the recovery stage, a sampled item may or may not be successfully recalled, with recovery probability determined by the absolute strength of the cue-item association.
This two-stage framework elegantly explains list-strength effects, part-list cuing inhibition, the difference between recall and recognition, and output interference (earlier recalled items change the sampling probabilities for later items). SAM was one of the first models to provide a unified mathematical account of both recall and recognition within a single framework, and it spawned the development of later models including REM and the retrieving effectively from memory framework.